Question

I have a string which contains some HTML code. I would like to find out if the HTML code represents visible text or an image. I used Java to solve this problem using the following regular expressions (I know you cannot parse HTML using RegExps, but I thought for what I am up to RegExps are enough).

public static String regex_html_tags_1 = "<\\s*br\\s*[/]?>";
public static String regex_html_tags_2 = "<\\s*([a-zA-Z0-9]+)\\s*([^=/>]+\\s*=\\s*[^/>]+\\s*)*\\s*/>"; 
public static String regex_html_tags_3 = "<\\s*([a-zA-Z0-9]+)\\s*([^=>]+\\s*=\\s*[^>]+\\s*)*\\s*>\\s*</\\s*\\1\\s*>"; 

public static String[] HTMLWhiteSpaces = {"&nbsp;", "&#160;"};

The code using these RegExps works fine for strings like

<h2></h2>

or alike. But a string

<img src="someImage.png"></img>

is also thought of as being empty.

Does anyone have a better idea than using RegExps to find out if some HTML code actually represents human readable text when it is interpreted by a browser? Or do you think my approach eventually leads to success?

Thanks a lot in advance.

Answer 1

Try using JSoup. It let’s you parse HTML documents using css selectors (jquery-style).

A very simple example to select all non-empty elements would be:

Document doc = Jsoup.connect("http://my.awesome.site.com").get();
Elements nonEmpties = doc.select(":not(:empty)");    

The full-blown solution will of course require some extra work to do, like

• iterating over lists of elements,
• checking css styles (for display or visibility or sizes or overlaying elements)
• checking src attributes for images
• etc

but it’s definitely worth it. You’ll learn a new framework, discover possibilities to ‘hide’ content in HTML / CSS and – most important – stop using regular expressions for HTML parsing 😉